How China’s WTO-era manufacturing surge reshaped world trade

Between its 2001 WTO accession and the mid-2010s, China’s share of world manufactures exports roughly quintupled. The shock propagated into prices, labour markets, and the complexity ranking of nations. Seven figures trace who gained productive knowledge, who lost manufacturing employment, and where the flows have moved since the 2018 US tariffs.

1. From 4% to 27% of world machinery in twenty-five years

Autor, Dorn & Hanson (2013, AER) date the China shock to the 1991–2007 window, with WTO accession in December 2001 the clean policy discontinuity. Pierce & Schott (2016, AER) identify the same break from the US conferral of permanent normal trade relations (PNTR), which eliminated the annual tariff-renewal uncertainty. The aggregate footprint is visible at the HS Section level: China’s share of world exports in five manufacturing sections rises from low single digits in 1995 to double digits by the late 2000s.

2. Diversification and the complexity upgrade

Hausmann, Hwang & Rodrik (2007, Journal of Economic Growth) argue that what a country exports matters for subsequent growth — not all dollars are equal. Felipe, Kumar & Abdon (2012) show that China’s post-WTO trajectory is exceptional: it simultaneously widened the basket (extensive margin) and shifted it toward more complex products (intensive margin). The first panel counts HS6 products where China has revealed comparative advantage (Balassa RCA ≥ 1); the second tracks the mean PCI of that revealed basket.

3. Chinese import penetration into OECD manufacturing

Autor, Dorn & Hanson (2013, AER 103(6): 2121–2168) define a local-labour-market exposure measure ΔIPW_jt = Σ_p (L_jp,1990 / L_j,1990) · (ΔM_China,p,t / L_p,1991), where L values come from 1990 commuting-zone and 1991 national employment. That 1990/1991 pre-shock base and the 1991 start-of- post-shock window are exactly what makes the instrument credibly exogenous. Acemoglu, Autor, Dorn, Hanson & Price (2016, Journal of Labor Economics 34(S1): S141–S198) aggregate those local estimates to 2.0–2.4 million US jobs displaced over 1999–2011. BACI starts in 1995, so we cannot replicate the 1991 employment base; we instead use 1995 manufacturing imports (the earliest pre-WTO snapshot BACI offers) as a dollar-denominated exposure base and divide by the 2000–2016 cumulative China bilateral flow. That ratio tracks Chinese penetration relative to a pre-accession import footprint; it is not ΔIPW and the two are not comparable in levels, because the denominators differ (worker-count vs dollar-of-imports) and the ADH numerator is an import change per worker, not a cumulative dollar flow. We cite ADH (2013) for the concept of China-import exposure, not the computation.

4. Value-per-ton trajectories for China-surge vs control HS6

One recurring conjecture in the China-shock literature is that a large low-cost supplier entering a product space compresses world prices for those products relative to the rest of the basket. We stay descriptive: index the median world export value-per-ton (USD/ton, from BACI) of the 200 HS6 products where China’s RCA rose fastest 2000–2010, and compare with the 200 products where it rose least. Value-per-ton is not a price — within each HS6 it mixes genuine price change with shifts in variety composition and weight-per-unit (a move from heavier to lighter varieties raises value-per-ton without any price change). The ADH-style selection on ΔRCA is exactly the set where China entered, which mechanically shifts the mass-weighted mix, so this is a composition-contaminated series, not the Broda-Weinstein (2006) new-variety import-price index. No welfare claim is drawn from it here.

5. Reallocation after the 2018 tariffs

Freund, Mulabdic & Ruta (2023, World Bank) and Alfaro & Chor (2023, NBER w31796) document that the Section 301 tariffs and subsequent US-China decoupling triggered third-country substitution rather than reshoring: Vietnam, Mexico, and smaller Asian producers captured the incremental flows. Alfaro & Chor use the actual Section 301 tariff schedules as the product list; we use a different, ex-ante basket — the 200 HS6 where China’s ΔRCA was largest over 2000–2010 — which is eight-plus years stale at 2017 and does not coincide with the tariffed set. What follows is therefore descriptive rather than a clean test of post-tariff substitution: it shows how producer shares in China’s 2000s-rise basket evolve through 2024, not whether that basket is the one being actively substituted in 2017–2024.

6. Complexity catch-up relative to peer economies

Hanson (2012, JEP “The Rise of Middle Kingdoms”) frames China’s post-WTO trajectory alongside other developing-country risers. The ECI is zero-centred: positive means above-median productive knowledge relative to the world, negative means below. We plot the trajectories of China, South Korea (benchmark “already-upgraded” tiger), Vietnam (follower), India (large-diversified-slow-moving), and Bangladesh (narrow-basket follower).

7. Bangladesh in the apparel slipstream

Mostafa & Klepper (2018, Organization Science) trace the emergence of the Bangladesh ready-made-garment industry to the late-1970s Daewoo-Desh technology transfer and the subsequent diffusion across domestic firms. The industry expanded first behind the Multifibre Arrangement, then accelerated after 2005 under the China-plus-one supplier diversification logic. We plot each country’s share of world apparel exports (HS chapters 61 & 62), 2000–2024.

8. The second China shock: EVs, solar, and lithium-ion batteries (2018-2024)

The first China shock was concentrated in HS chapters 61–63 (apparel), 84–85 (machinery and consumer electronics), and the lower-tech end of the complexity ladder. The post-2018 trajectory looks different. Autor, Beaumont, Dorn, Hanson & Pelgrom (2025, NBER w33375) and Alfaro & Chor (2024, JEP) call the surge in Chinese exports of electric vehicles (HS 870380), solar modules (HS 854143), and lithium-ion batteries (HS 850760) the “second China shock”. It is concentrated in green-transition capital goods, it comes after the Section 301 tariffs were already in force, and it has prompted the IRA (US 2022) and the EU Net-Zero Industry Act (2024) anti-subsidy responses. We read China export levels for each HS6 in 2018 and 2024 and compute China’s share of world exports in each year.

9. Second-shock update, 2023-24: where the flagship products landed

The 2018 tariffs rewired where China’s EV, solar, and battery exports went, not whether they grew. Using BACI 202501 HS6-extension data for 2023 and 2024, we trace the destination mix for the three flagship second-shock products. USA is effectively closed to Chinese EVs under Section 301 (raised to 100% ad valorem in 2024, USTR proclamation of 14 May 2024) and to Chinese solar modules through UFLPA enforcement plus anti-dumping/CVD orders; Germany and Japan are the two largest open advanced-economy destinations. The residual “rest of world” bar captures the emerging-market destinations — Southeast Asia, Latin America, the Gulf — where no comparable trade-defence instrument is in force. Alfaro & Chor (2024, JEP) frame this as “decoupling by destination”: the same Chinese product space gets split across markets depending on the local policy response.

10. Where did China’s share gain come from? Bloc-level share-shift, 1995 vs 2024

Figure 1 shows the share China gained. The mirror question, who lost it, is the share-shift decomposition Hanson (2012, JEP 26(4): 41–64) frames as the defining empirical content of the “Rise of Middle Kingdoms”. For each of the five manufacturing HS sections we partition world exports into four mutually exclusive blocs — China, the G7 ex-Japan (USA, GBR, DEU, FRA, ITA, CAN), Japan, and the Asian followers (KOR, VNM, IDN, MYS, THA, PHL, BGD, IND) — and compute each bloc’s share in 1995 and 2024. The change in each bloc’s share is the per-section share lost or gained over the window. By construction, the bloc deltas sum to the residual ROW shift.

Eight panels, one story

China’s share of world manufactures exports rose because the revealed basket widened and climbed the complexity ladder; prices of targeted products compressed relative to the rest of the basket; OECD economies with exposed manufacturing sectors absorbed a cumulative demand shock on the Autor-Dorn-Hanson scale; the 2018 US tariffs redirected flows through Vietnam, Mexico and India rather than reversing Chinese dominance; follower economies from Vietnam to Bangladesh are capturing market shares in product ladders China is vacating as wages rise; and a second China shock in electric vehicles, solar modules, and lithium-ion batteries is now landing on the same OECD economies that absorbed the first.

Units: trade values current USD (BACI stored in thousands, ×1,000 for display). Unit values USD/ton. PCI and ECI are σ from world mean, zero-centred — negative values routine. BACI-numeric codes: CHN=156, USA=842, KOR=410, VNM=704, IND=699, BGD=50, IDN=360, MEX=484. Apparel = HS 61+62. Manufacturing = HS 28–96.